£40 METALLURGY OF IRON AND STEEL. out; and Table XIII-0 gives investigations made under my supervision. Under another system of plate rolling, practiced at the larger American mills., and extensively abroad, it is the practice to make larger ingots which are rolled into slabs, these being reheated for the plate train. It would be supposed that these slabs would show greater segregation than is found in plate ingots, but this assumption is hardly sustained by Table XIII-D, which gives the results obtained by drilling into the axial line of slabs rolled from 'large ingots,, made by The Pennsylvania Steel Co. The paints ; helow the top crop end, and one-third way down the ingot, include :the most contaminated region. The concentration in these cases probably marks the "extent of the action of simple crystallization, • while more extreme cases would represent the liquation of fusible * impure compounds. : SEC. XHId.—Homogeneity in plates.—The fact that plates are not homogeneous when rolled from ordinary ingots does not become evident under ordinary inspection, since, generally, only one test-piece is taken from the sheet, and this comes from the edge, but it will be shown by Table XIII-B that the variations are by no means unimportant. The first instance is from Pourcel,* the next three from Cunningham,! while the last two are from my own investigations. The data on heat 11,393 were obtained by rollr ing an ingot on a universal mill into a long plate. The upper third of this plate was sheared into 16-inch lengths, and tests taken along the center line and the edge. A strip was also cut from the bottom end of the plate in the center and on the edge. The tests of heat 10,768 were from a "pitted" plate. The flaws in the -bars render worthless any records of elongation, but the chemical results are valuable, while the determinations of tensile strength are approximately correct. The ingot was rolled on a shear mill to a thickness of three-quarter inch. • The plate was only 112 inches ! long after trimming, so that the seven tests represent" the entire length of the sheet. A great deal of this irregularity between different parts of the j same plate may be avoided by rolling from a slab. It would be untrue to say that segregation can be avoided by making a larger *ioc. cit. t Trans. A.I.M.E., XXIII, p, 626, et seq.